Pages: 9-19

Abstract

Banking system fraud is one of the challenges of banking and e-commerce development. One of the main challenges of machine learning and data mining techniques in bank fraud detection is their low accuracy in identifying these transactions. This research presented a hybrid method based on a multi-layer artificial neural network and bat algorithm to reduce the fraud detection fault. In the proposed method, the parameters of the neural network such as weights and bias are selected optimally by the bat algorithm to reduce the fault rate in the fraud detection. The proposed method is a type of learning intensification in which the bat algorithm improves the learning of the neural network. In the proposed method, the Kmeans clustering is used to remove data from the dataset to increase the accuracy of the proposed method. MATLAB software was used to run the data. The data related to bank fraud indicate that the accuracy, sensitivity, and specificity of the proposed method for detecting bank fraud were as much as 91.46%, 88.97%, and 90.32%, respectively. The comparison of our proposed method with other methods shows that the proposed method is more accurate than methods such as regression and backup machine